Variation in the Eastern Ocean throughout the whole voyage to Goa and the Moluccas is observed by the Portuguese; but they err greatly in many things, following, as they do, the first observers who note down variations in certain places with ill-adapted instruments, and by no means accurate observations, or by some conjectures. As, for instance, in Brandöe Island, they make the versorium deviate by 22 degrees to the north-west. For in no region or place in the whole world, of not greater latitude, is there so great a deviation; and, in reality, there the deviation is slight. Also when they make out that at Mosambique the compass deviates by one rumbe to the north-west, it is false; even though they use (as they are accustomed to do) the Portuguese compass: for beyond all doubt on the shore of {182} Mosambique the versorium inclines ¼ rumbe or even more to the south-west. Very wrongly also beyond the æquator in the course to Goa they make the little compass incline by 1½ rumbe to the west: whereas they should rather have said that in the first part of the course the Portuguese compass inclines by 1 rumbe: but that the true meridional compass inclines by ½ rumbe only. In order that the amount of variation in the Eastern Ocean may be accurately settled in most places by our rules, there is needed a more exact and truer survey of the southern land, which spreads out from the south to the æquinoctial more than is commonly described on maps and globes. * * * * * CHAP. XXI. How the deviation of the versorium is augmented and _diminished by reason of the distance of places_. In the middle of great and continent lands there is no variation. Nor, generally, in the middle of very great seas. On the margin of those lands and seas the variation is often ample, yet not so great as at a little further distance on the sea. As, for example, near Cape St. Augustine the compass varies; but at 50 miles from land toward the East it varies more; and 80 miles off it varies still more; and yet still more at a distance of 100 miles. But from a distance of 100 miles the diminutions of deviation are slower, when they are navigating toward the mainland, than at a distance of 80 miles, and at a distance of 80 miles than at 50: for the deviations change and are diminished rather more swiftly the more they approach and draw near land than when at a great distance off. As, for instance, navigating toward Newfoundland the change of variation is more rapid (that is, it decreases a degree in a smaller arc of the course on the parallel) when they are not far from land than when they are a hundred miles distant: but when travelling on land toward the interiors of regions the changes are slower in the first parts of the journey than when they come more into the interior.